Discovery of a series of imidazopyrazine small molecule inhibitors of the kinase MAPKAPK5, that show activity using in vitro and in vivo models of rheumatoid arthritis

Article abstract of DOI:10.1016/j.bmcl.2012.01.077

MAPKAPK5 has been proposed to play a role in regulation of matrix metalloprotease expression and so to be a potential target for intervention in rheumatoid arthritis. We present here the identification of a series of compounds against this target which are effective in both biochemical and cell assays. The expansion of the series is described, along with early SAR and pharmacokinetics for some representative compounds.

Full text of DOI:10.1016/j.bmcl.2012.01.077

Bioorganic & Medicinal Chemistry Letters 22 (2012) 2266–2270
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of a series of imidazopyrazine small molecule inhibitors of the kinase
MAPKAPK5, that show activity using in vitro and in vivo models of
rheumatoid arthritis
Martin J.I. Andrews a, , J. Andrew Clase a, Gregory Bar a, Giovanni Tricarico a, Paul J. Edwards a,à
,
⇑
Reginald Brys a, Mark Chambers b, Wolfgang Schmidt b, Angus MacLeod b, Kim Hirst b, Vivienne Allen b,
Veronique Birault b, , Joelle Le b, , John Harris b, Andrew Self b, Kevin Nash b, Graham Dixon a
a Galapagos NV, Generaal de Wittelaan L11 A3, 2800 Mechelen, Belgium
b BioFocus, Chesterford Research Park, Saffron Walden, Essex CB10 1XL, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
MAPKAPK5 has been proposed to play a role in regulation of matrix metalloprotease expression and so to
be a potential target for intervention in rheumatoid arthritis. We present here the identification of a ser-
ies of compounds against this target which are effective in both biochemical and cell assays. The expan-
sion of the series is described, along with early SAR and pharmacokinetics for some representative
compounds.
Received 20 December 2011
Revised 20 January 2012
Accepted 21 January 2012
Available online 31 January 2012
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
MAPKAPK5
Rheumatoid arthritis
Kinase
MMP
Rheumatoid arthritis (RA) is a chronic autoimmune inflamma-
tory and joint degenerative disease that affects almost 1% of the
adult population worldwide, with onset classically between the
ages of 30 and 50 and higher prevalence in women.1 Clinically,
RA is characterised by relapse/remitting symmetric inflammation
of multiple joints that ultimately results in irreversible damage
of the articular cartilage and in erosion of adjacent bone.2 When
left untreated, the disease leads to significant disability associated
with high economic costs, and even premature death.
The current frontline therapy for this disease is the disease-
modifying anti-rheumatic drug (DMARD) methotrexate (MTX)
which, while useful in the early stages, is less effective as the
disease progresses, where steroid treatment is typically used to re-
duce acute flares of the disease.3 This approach is now supported
by biological therapies such as EnbrelÒ, RemicadeÒ, HumiraÒ,
Rituximab and Abatacept that work by suppression of the cytokine
signalling that drives the disease.4 However, due to the high costs
and inconvenience of parenteral administration of biologics and to
many patients who only partially responding to biological thera-
pies, there is still a need for less expensive and orally administered
drugs with a novel mode of action. Hence, the development of a
number of small-molecule inhibitors targeting disease-relevant
signal transduction pathways is being actively pursued by various
companies.5
We recently identified a new target for therapeutic intervention
in RA, MAPKAPK5, by screening disease-relevant cells (synovial
fibroblasts from rheumatoid arthritis patients, RASFs) for reduction
of expression of MMP1, a joint degrading metalloprotease, on stim-
ulation with cytokines.6 As both the cell type, derived from the in-
flamed synovium (‘pannus’), and trigger are accepted as playing a
key role in RA pathogenesis, this assay is thought highly disease-
relevant. MAPKAPK5 (mitogen-activated protein kinase activated
protein kinase 5), also known as p38-regulated/activated kinase
(PRAK), is structurally related to other known serine/threonine
kinases that function downstream of MAPKs, is expressed in all
tissues and is activated in response to cellular stress and pro-
inflammatory cytokines. At least part of this activation is mediated
⇑
Corresponding author. Tel.: +32 15 342 940; fax: +32 15342 901.
E-mail address: martin.andrews@glpg.com (M.J.I. Andrews).
Present address: GlaxoSmithKline R&D Ltd, Medicines Research Centre,
Stevenage, UK.
through p38
exact role of p38
a
(MAPK14), but some controversy exists over the
a
in the activation of MAPKAPK5.7 However, the
physiological role and downstream targets of MAPKAPK5 remain
largely undefined.
à
Present address: Department of Chemistry, Boehringer Ingelheim (Canada) Ltd,
2100 Cunard Street, Laval, Quebec, Canada H7S 2G5.
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2012.01.077

M. J. I. Andrews et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2266–2270
2267
performed in ADME assays including microsomal stability and sol-
ubility, and Series 1 was selected as the best starting point for hit-
to-lead development based on overall profile.
For the synthesis of compounds from Series 1, the dibromoim-
idazopyrazine 7 was prepared as shown in Scheme 1. 3-Aminopyr-
azine-2-carboxylic acid 1 was esterified with methanol in the
presence of H2SO4, and the resulting pyrazine ester 2 brominated
in the 6-position using Br2 in AcOH to furnish intermediate 3.
The 3-bromo substituent was introduced using a Sandmeyer
reaction affording intermediate 4, which was converted into the
2-aminopyrazole precursor 6 by hydrolysis followed by Curtius
rearrangement. Condensation of
6 with bromoacetaldehyde
diethyl acetal provided dibromoimidazopyrazine 7.
Initial nucleophilic substitution to introduce the R1 group as de-
fined in 9 using an aromatic amine was found to be possible under
basic conditions; however for electron-deficient amines it was
found that reactions were extremely slow. In these cases it was
found that acidic catalysis conditions could also be applied to in-
crease reaction rate. This reaction was also found to be regioselec-
tive, leaving the C8 bromine available for subsequent palladium
catalysed coupling to introduce the R2 group. In general, Suzuki
couplings9 using either boronic acids or boronic esters were uti-
lised, and a number of conditions were established for different
types of boronate. However, where required, Stille couplings10
were also used; indeed, it was found that stannylation of the core11
was possible to increase the range of groups that could be coupled;
boronation of the core did not prove possible.
Initial screening showed that an aromatic group at R1 was
favourable for potency, since the library compounds containing
an alkyl or benzyl substituent at this position had low activity
against MAPKAPK5. Methylation of this amine was also found to
suppress activity greatly (10 vs 11), as was substitution of the
amine by an ether linker (data not shown), indicating the critical
importance of a secondary amine in this position. The nature of
the aromatic ring and substitutions were found to be less critical
for activity.
Figure 1. Overview of initial hit series.
Following target discovery, which will be discussed elsewhere,
a functional assay using radioactive ATP was created8 and used
for screening of the BioFocus SoftFocusÒ Kinase collection, which
at the time comprised approximately 16,000 compounds. Primary
screening in the functional kinase assay for MAPKAPK5 gave 468
hit compounds, a hit rate of 2.8%. Full IC50 data were then gener-
ated for the 100 compounds with the highest percentage inhibi-
tion. The active compounds were represented by three main
series shown in Figure 1. Several representative compounds from
each of these series were then profiled further, including screening
in the RASF primary cell assay used for target identification, using
compounds in place of target knockdown. Further testing was also
The SAR was then developed further in the series as shown in
Table 1. This demonstrated that the relative freedom for
Scheme 1. General synthetic scheme for production of compounds. (i) H2SO4, MeOH; (ii) AcOH, Br2; (iii) NaNO2, HBr, H2O; (iv) LiOH, MeOH, H2O, (v) DPPA, t-BuOH, TEA reflux
then TFA/DCM (vi) bromoacetaldehyde diethyl acetal, iPrOH, (vii) NH2-R1, iPrOH, (viii) R2B(OR), Pd(OAc)2, PPh3.

2268
M. J. I. Andrews et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2266–2270
Table 1
SAR around R2 group in kinase and phenotypic assay
Compound
R2 group
MAPKAPK5 IC50 (nM)
829
>10000
1490
1370
>10000
2390
390
>10000
>10000
1420
2820
3200
437
MMP1 cell IC50 (nM)
2990
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
4-Pyrazolyl
(7-N-methyl) 4-pyrazolyl
4-Triazolyl
>10000
914
>10000
>10000
871
4-Benzenecarboxamido
4-Benzosulphonamido
4-(3-Fluoro)benzamide
4-(2-Pyridone)
4-(2-Ethoxypyridine)
4-(2-Dimethylaminopyridine
2-Thienyl-5-carboxamide
2-Thiazolyl-5-carboxamide
5-Thiazolyl-2-carboxamide
3-Thienyl-5-carboxamide
5-Isoindolin-1-one
392
>10000
>10000
327
239
>10000
148
676
5-Indole
2-Furyl-4-carboxamide
3-Furyl-5-carboxamide
227
702
88
modification in R1 was not matched at R2, where far tighter struc-
tural requirements were evident, as shown by comparisons 16–13,
16–17 and 23–13. The biochemical assay results showed clearly
that both an H-bond acceptor and donor element had to be present
for optimum activity (16 vs 17, 18). Systems with only one donor
or acceptor had significantly lower activity. Amongst the other
groups found to be effective was the 2-pyridone 16, where the
amide is contracted into the ring. This group showed activity
equivalent or superior to the benzamide 13, indicating that the dis-
tance from hinge binding unit to the pendent amide may be impor-
tant for activity. This perhaps indicates a degree of protein
flexibility in this area, which is further supported by the activity
of the pyrazole compound 10, where the drop in activity is rela-
tively low, despite being markedly smaller in terms of steric bulk.
This was further supported by incorporation of alternative hetero-
cyclic amides with correspondingly different electronic and steric
characteristics (19–23, 25, 26).
In general, amide groups have been associated with poor ADME
properties, especially in terms of the Caco2 permeability assay.12
However, no general issue was found with the series, as com-
pounds bearing these groups did not show issues related to poor
permeability and exposure. No specific issues were detected with
the core structure in regard to microsomal stability assays, with
the results being clearly dependant on the nature of side chains.
In addition, screening of a number of analogues in both CYP450
inhibition and hERG binding assays showed no core-specific liabil-
ities for the series.
Figure 2. Docking of 10 in active site of MAPKAPK5 showing hydrogen bond
network with hinge region.
Compounds of the series were also tested in RA related pheno-
typic cellular assays, where the inhibition of MMP1 and MMP13
production was clearly observed. As these proteases are strongly
linked to cartilage destruction, we took the view that inhibition
of the target was leading to a down regulation of these markers
of disease progression. These results correlated well with our tar-
get- finding exercise, which showed similar reductions in MMP1
levels to those which had been achieved using knock-down tech-
nologies including SilenceSelect adenoviral libraries and siRNA.13
An understanding of the protein-ligand interactions would en-
able the rational design of further compounds within the series,
and potentially allow a scaffold-hopping exercise to be performed.
Since no structural data were available for MAPKAPK5, a homology
model was built, based on the closest homologue identified,
MAPKAPK2.14 MAPKAPK2 shares a 42% sequence identity with
the kinase domain of MAPKAPK5. Of the three MAPKAPK2 crystal
structures available at the time, 1NXK15 was selected as the
template structure, due to the inclusion of a ligand and the best

M. J. I. Andrews et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2266–2270
2269
Table 2
Comparison of Series 1 and 2 compound results
R2 group
Series 1
Series 2
Compound
MAPKAPK5 IC50 (nM)
MMP1 cell IC50 (nM)
Compound
MAPKAPK5 IC50 (nM)
MMP1 cell IC50 (nM)
4-Pyrazolyl, X = O
5-Isoindolin-1-one, X = O
4-(2-Pyridonyl), X = NMe
10
23
27
829
239
290
2823
858
871
28
29
30
816
195
3730
2608
445
NA
resolution available. A selection of active compounds were then
docked into the MAPKAPK5 model, resulting in the identification
of numerous potential binding modes.16 The most likely binding
mode was then selected by review of the structures against the
available SAR (Fig. 2). The predicted binding mode showed a bi-
dentate hydrogen bond pattern at the hinge involving the imidaz-
ole nitrogen and aryl amine, in complete agreement with the SAR,
for example comparison of compounds 10 and 11. However, the
requirement for a hydrogen bond donor and acceptor pair in the li-
gand at the 5-position was not fully understood from this model of
binding. It was postulated that water-mediated interactions,
including those between the conserved lysine and aspartic acid
residues and others in the phosphate binding sub-pocket may be
responsible, however no direct interactions were found by model-
ling. The substituent at the 8-position was predicted to be oriented
toward the bulk solvent, again in close agreement with the ob-
served SAR, where greater tolerance for substitution had been pre-
viously noted from the primary screening output.
With the SAR and an understanding of the binding mode in
hand, we undertook a scaffold hopping exercise, aimed at further
optimisation of the interactions of the core with its target and to
provide backup to the current series. A number of related, novel
templates with the potential to retain the same H-bond interac-
tions were designed and then assessed for similarity to the
electronic character found in the starting reference compound,
along with likely synthetic accessibility. The list was reduced from
around 40 to 9 cores using these methods, and a number of
compounds from each of these cores were then synthesized and
tested for activity. Results for one of these structural types are
shown in Table 2, detailing compounds that showed equivalent
activity in the kinase and cell assays. Interestingly, several core
structures showed significant loss of activity against MAPKAPK5,
even where equivalent multiple H-bond networks could still be
formed (data not shown). This reflects the hypothesis that tuning
of individual H-bond strengths can be used as a method to drive
selectivity from the hinge interaction, in addition to the more reg-
ularly used tools of engineering steric and electronic clashes. In
addition, it was noted that transfer of SAR between series was
not complete, with some groups markedly poorer in Series 2 than
in Series 1, for example, 27 versus 30. Docking of these compounds
in MAPKAPK5 did not show significant differences in binding
mode, so it is unclear what drives the distinct activity seen in the
assay.
The initial compound (28) found in Series 2 demonstrated com-
parable activity in the MAPKAPK5 assay, and also in the phenotypic
cell assays. Additional testing was also performed in microsomal
stability, solubility, CYP450 inhibition and Caco2 assays, and
gratifyingly it was found that few fundamental liabilities existed,
as had also been observed in Series 1. In fact, where certain Series
1 compounds had demonstrated poor Caco2 assay data, this issue
was not seen in Series 2. These characteristics resulted in good
PK properties being found for both the new Series 2 as well as Ser-
ies 1 (Fig. 3), and so compounds from the series were moved for-
ward to in vivo models of RA, which will be reported later.
In summary, we have generated a screen to detect inhibitors of
MAPKAPK5, and screening 1611 compounds provided hit Series 1,
Figure 3. PO exposure (rat, 5 mgkgÀ1, 0.5% methylcellulose + 1% DMSO) for 23 and 29.

2270
M. J. I. Andrews et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2266–2270
4. (a) McInnes, I. B.; O’Dell, J. R. Ann. Rheum. Dis. 1898, 2010, 69; (b) van
Vollenhoven, R. F. Nat. Rev. Rheumatol. 2009, 5, 531.
5. (a) Cohen, S.; Fleischmann, R. Curr. Opin. Rheumatol. 2010, 22, 330; (b) Opar, A.
Nat. Rev. Drug Disc. 2010, 9, 257.
6. Andrews, M. J. I.; Brys, R.; Vandeghinste, N.; Pujuguet, P.; Namour, F.; Lorenzon,
G.; Chambers, M.; Schmidt, W.; Clase, J. A.; Birault, V.; Dixon, G. Ann. Rheum.
Dis. 2009, 68, 590.
7. (a) New, L.; Jiang, Y.; Zhao, M.; Liu, K.; Zhu, W.; Flood, L. J.; Kato, Y.; Parry, G. C.
N.; Han, J. EMBO 1998, 17, 3372; (b) Seternes, O.-M.; Mikalsen, T.; Johansen, B.;
Michaelsen, E.; Armstrong, C. G.; Morrice, N. A.; Turgeon, B.; Meloche, S.;
Moens, U.; Keyse, S. M. EMBO 2004, 23, 4780; (c) New, L.; Jiang, Y.; Han, J. Mol.
Biol. Cell 2003, 14, 2603.
which possessed good biochemical and cellular activities but also
some liabilities that were addressed by performing a scaffold
hopping exercise that led to a novel series of kinase inhibitors.
On the basis of these results, a new Series 2 was then taken
forward as the basis for a lead optimisation project which resulted
in the generation of a compound which reached a phase II clinical
trial.
Supplementary data
8. MAPKAPK5 protein was supplied in active form by the University of Dundee.
9. (a) Suzuki, A. Pure Appl. Chem. 1991, 63, 419; (b) Andrews, M. J. I.; Edwards, P. J.;
Chambers, M. S.; Schmidt, W.; Clase, J. A.; Bar, G.; Hirst, K. L.; Macleod, A. WO
2007138072.
10. Stille, J. K.; Echavarren, A. M.; Williams, R. M.; Hendrix, J. A. Org. Synth., Coll.
1998, 9, 553.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2012.01.077. These data
include MOL files and InChiKeys of the most important compounds
described in this article.
11. Nierengarten, H.; Rojo, J.; Leize, E.; Lehn, J.-M.; van Dorsselaer, A. Eur. J. Inorg.
Chem. 2002, 573.
12. van Breemen, R. B.; Li, Y. Expert Opin. Drug Metab. Toxicol. 2005, 1, 175.
13. The procedure used for compound testing was to add compounds to RASFs for
References and notes
at least 30 min before addition of the TNF
a trigger and incubation for 48 h.
1. Mount, C.; Featherstone, J. Nat. Rev. Drug Disc. 2005, 1, 11.
2. (a) Smolen, J. S.; Steiner, G. Nat. Rev. Drug Disc. 2003, 2, 473; (b) Sweeney, S. E.;
Firestein, G. S. Int. J. Biochem. Cell Biol. 2004, 36, 372; (c) Firestein, G. S. Nature
2003, 423, 356.
3. Smolen, J. S.; Landewé, R.; Breedveld, F. C.; Dougados, M.; Emery, P.; Gaujoux-
Viala, C.; Gorter, S.; Knevel, R.; Nam, J.; Schoels, M.; Aletaha, D.; Buch, M.;
Gossec, L.; Huizinga, T.; Bijlsma, J. W.; Burmester, G.; Combe, B.; Cutolo, M.;
Gabay, C.; Gomez-Reino, J.; Kouloumas, M.; Kvien, T. K.; Martin-Mola, E.;
McInnes, I.; Pavelka, K.; van Riel, P.; Scholte, M.; Scott, D. L.; Sokka, T.; Valesini,
G.; van Vollenhoven, R.; Winthrop, K. L.; Wong, J.; Zink, A.; van der Heijde, D.
Ann. Rheum. Dis. 2010, 69, 964.
Supernatant was then collected and subjected to a MMP1 ELISA developed in
house. IC50 values were derived from concentration–response plots.
14. Tripos version 7.1.
15. Underwood, K. W.; Parris, K. D.; Federico, E.; Mosyak, L.; Czerwinski, R. M.;
Shane, T.; Taylor, M.; Svenson, K.; Liu, Y.; Hsiao, C.-L.; Wolfrom, S.; Maguire, M.;
Malakian, K.; Telliez, J.-B.; Lin, L.-L.; Kriz, R. W.; Seehra, J.; Somers, W. S.; Stahl,
M. L. Structure 2003, 11, 627.
16. Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, A. G. J. Mol. Biol. 1996, 261, 470.
FlexX version 1.